See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/318014600

Dynamic and residual behavior of reinforced

concrete floors following instantaneous
removal of a column

Article in Engineering Structures · October 2017

DOI: 10.1016/j.engstruct.2017.06.059

CITATIONS READS

0 11

2 authors:

Qian Kai Bing Li

Guangxi University Nanyang Technological University
36 PUBLICATIONS 246 CITATIONS 137 PUBLICATIONS 1,086 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Bing Li on 29 June 2017.

The user has requested enhancement of the downloaded file.

 Engineering Structures 148 (2017) 175–184

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Dynamic and residual behavior of reinforced concrete floors following

instantaneous removal of a column
Kai Qian a, Bing Li b,⇑
a
College of Civil Engineering and Architecture, Guangxi Key Laboratory of Disaster Prevention and Engineering Safety, Guangxi University, 100 Daxue Road, 530004, China
b
School of Civil and Environmental Engineering at Nanyang Technological University, 639798, Singapore

a r t i c l e i n f o a b s t r a c t

Article history: With the increasingly real threat of terrorism around the world, many iconic and important buildings
Received 14 March 2016 may become targets of terrorist attack. This can lead to severe casualties and economic losses as such
Revised 17 May 2017 extreme loading events are not considered in conventional building design. However, experience from
Accepted 23 June 2017
past events has shown that most well designed structures will not completely collapse immediately upon
the loss of one or more columns as in cases of sabotage. As such, the residual load resisting capacity is
crucial in post damage situations where engineers have to make a decision of whether to rehabilitate
Keywords:
or rebuild the damaged building. To better understand the behavior of such damaged structures, a series
Progressive collapse
Reinforced concrete
of tests are conducted in this present study. The dynamic response of the specimens is assessed through
Three-dimensional substructures dynamic tests. Following dynamic tests, the specimens, which have suffered different degrees of damage,
Dynamic response are re-tested by push-down loading regimes to capture their residual behavior. The experimental and
Residual behavior analytical results indicated that the damage caused by the dynamic response will significantly degrade
the initial stiffness and detriment the efficiency of compressive arch action and compressive membrane
action even if the specimens actually only experience elastically dynamic response. However, when the
specimens undergo a considerable plastically dynamic response, no compressive arch action and com-
pressive membrane action are able to develop. The load resisting capacity would derive mainly from ten-
sile membrane action and catenary action in large deformation stage.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction A 3/8 scaled RC beam-column sub-assemblage (Sasani and Kro-

Past events such as bomb attacks on the Murrah Federal Build-
ing, 1995, Khobar Towers, Saudi Arabia, 1996, London Docklands,
1996, World Trade Center, 2001, and Lahore City, Pakistan, 2008
have shown that intentional targeted bomb attacks by terrorists
may result in severe loss of live and properties. Researchers (Sasani
and Kropelnicki [1]; Sasani et al. [2]) have shown that the main
cause of severe casualties is the occurrence of progressive collapse
rather than direct blast pressure, such as the collapse of Murrah
Federal Buildings in 1995. Thus, it is necessary to evaluate the
capability of the buildings to mitigate progressive collapse. Among
various methods, the Alternate Load Path (ALP) method is fre-
quently utilized to investigate the ability of the buildings to redis-
tribute the loads safely from the notionally removed column to
remaining surrounding elements. A number of researchers have
conducted experimental investigations on the behavior of rein-
forced concrete (RC) buildings to resist progressive collapse via
ALP method.
⇑ Corresponding author.
E-mail address: cbli@ntu.edu.sg (B. Li).
pelnicki [1]) was tested via push-down loading method (a typical
expression of ALP method) to investigate the behavior of RC frames
subjected to the loss of an interior or exterior column scenario. It
concluded that catenary action developed in top reinforcement is
the main source of the load resisting capacity in large displacement
stage. The catenary action is lost when the beam end rotation
reaches 11 degrees.
Su et al. [3] conducted three series of reduced-scale tests to
investigate the influences of design parameters on compressive
arch action developing in RC beam-column sub-assemblages. It
was found that compressive arch action could contribute 50 to
169% extra load resisting capacity beyond the flexural strength of
the beams. Moreover, on applying fast loading, compressive arch
action still could be developed.
Qian and Li [4] tested two series of specimens to quantify the
slab effects on the behavior of RC buildings to mitigate progressive
collapse. It was concluded that the RC slab could increase the load
resisting capacity of the bare frame by 40% to 63% and reduce the
progressive collapse risks effectively.

http://dx.doi.org/10.1016/j.engstruct.2017.06.059
0141-0296/Ó 2017 Elsevier Ltd. All rights reserved.
176 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184

Qian and Li [5] studied the behavior of RC flat slab or flat plate was directly tested under push-down loading regime, was taken
subjected to the loss of a corner column scenario where punching as a control specimen. The remaining two specimens, which have
shear failure was identified as the most critical failure mode. How- similar dimensions and reinforcement details, were tested dynam-
ever, the experimental results indicated that drop panels could ically to simulate the specimens subjected to sudden column
improve the behavior of flat plate structures effectively and ensure removal scenarios. Following the dynamic tests, these two speci-
tensile membrane action to develop in the slab. mens with different degrees of dynamic damage were re-tested
As mentioned above, a number of experimental and analytical by push-down loading regimes to assess their residual behavior.
studies [1–10] had been carried out to understand the reliable load
resisting mechanisms for RC buildings to mitigate progressive col-
2. Description of test program
lapse. However, quasi-static push-down tests were applied in these
experimental studies. The realistic behavior of RC buildings sub-
2.1. Experimental specimens
jected to the sudden column missing scenarios could not be well
captured by push-down tests. As such, some dynamic tests, which
Three one-quarter scaled slab-beam-column RC specimens
are closer to reality, are also carried out in the past several years.
were tested in this study. These three specimens had identical
Sasani et al. [2] carried out an in-situ dynamic test on an actual
designed dimensions and reinforcement details. The prototype
10-story RC building following the explosion of an exterior column.
building of these three specimens is a four-story RC frame, which
As the column was not removed clearly and the live loads on the
was designed by Defence Science & Technology Agency (DSTA),
floor were removed before tests, the recorded maximum vertical
Singapore based on Singapore Code CP65 [15] and checked to sat-
displacement was only 6.4 mm and the building only achieved
isfy the requirements of ACI 318-08 [16]. The design live load (LL)
elastic response.
of the prototype specimen is 5.2 kPa. The dead load (DL) due to the
Sheffield et al. [11] conducted two dynamic tests on a special
RC slab with thickness of 254 mm is 6.4 kPa. The dimensions and
fabricated full-scale four-story RC building. Comparing to the test
reinforcement details of the typical specimen are shown in Fig. 1
carried by Sasani et al. [2], the additional dead load and design live
and Tables 1 and 2. The specimen was supported by five columns
load was applied before the removal of the columns. In addition,
with size of 170 mm
170 mm. The slab was extended by
the columns were removed clearly by pre-installed explosives. It
563 mm (refer to hatched zone in Fig. 1) to simulate the additional
was found that the maximum vertical displacement was 200 mm
constraints from the interior span adjacent to the test specimen.
when only one exterior column was removed. When the adjacent
The slab thickness is 64 mm with a concrete cover of 7 mm. As
interior column was also removed, the maximum vertical displace-
the cover is only 7 mm, chipping with maximum size of aggregate
ment could reach 968 mm with significant concrete crushing as
of 10 mm was used. As mentioned above, the Control Specimen
cracks formed in the beams and floors.
Con-1 was tested subjected to a push-down loading regime. The
As conducting full-scale multi-story tests is discouragingly
remaining two specimens were tested subjected to simulated sud-
expensive and it is very difficult to install critical instrumentations,
den column removal scenarios. One of the specimens was tested
simplified dynamic tests could be carried out in a controlled simu-
dynamically with externally applied pressure of 8.9 kPa. Thus,
lated environment instead. Qian and Li [12] conducted a series of
including the dead load of the slab with 64 mm thickness
1/3-scale dynamic tests with special designed Instantaneous Col-
(1.6 kPa), the total pressure applied on the specimen is 10.5 kPa,
umn Removal Devices (ICRD). The load cell installed beneath the
which is about 0.91(DL + LL) of the prototype frame. Thus, this
ICRD indicated that the ICRD could remove the temporary support
specimen is designated as D-0.91. Similarly, another dynamically
as fast as 0.005 s, which was fast enough to simulate the column
tested specimen is called D-1.16, as the externally applied pressure
removal due to extreme loading, such as bomb or vehicular impact.
is 11.9 kPa. As after suddenly removal of the column, both D-0.91
Thus, the ICRD will also be utilized in this study to simulate sudden
and D-1.16 are able to stabilize in some position and without com-
column loss scenarios.
plete collapse, their residual strength was measured by subsequent
When a building is subjected to the sudden column removal
push-down tests. In subsequent tests, these two specimens with
scenario, it may experience two phases: (a) dynamic phase, (b)
initial dynamic damage were called R-0.91 and R-1.16,
post-dynamic phase or called residual phase. If a push-down test
respectively.
based on intact specimen is taken as a study on the residual behav-
The target compressive strength of the concrete is 25 MPa while
ior, it will over-estimate its residual behavior and will result in
the measured compressive strength of Con-1, D-0.91, and D-1.16
unsafe design as the initial damage caused by the dynamic vibra-
are 23.0 MPa, 24.4 MPa, and 25.1 MPa, respectively. Five types of
tion was ignored. Existing push-down tests mainly focus on evalu-
reinforcing steel were utilized for reinforcing cage: R3, R6, R8,
ation of dynamic behavior and most researchers treat it as an
T10, and T13. The properties of reinforcing steel are tabulated in
alternative to complex dynamic tests. The dynamic increase factor
Table 3. For example, R3 and T10 represent plain round bar and
and analytical models [energy based model from Izzuddin et al.
deformed bar with diameter of 3 mm and 10 mm, respectively.
[13] and Single-Degree-Freedom-Model from Qian and Li [14]]
could convert the quasi-static response to dynamic response easily,
although their reliability still needs further confirmation. Although 2.2. Design of test setup
some buildings may totally collapse when several columns are
removed suddenly, many others with sufficient continuity, ductil- Fig. 2 shows the experimental setup of dynamic test D-0.91.
ity, and/or redundancy could survive (Sasani et al. [2]; Sheffield Five intact columns are fixed on the steel supports (Item 7 in
et al. [11]). However, the Dead Load (DL) and Live Load (LL) applied Fig. 2) by bolts while the notionally removed column is supported
on the floor will not decrease after removal of the columns. Thus, by the special designed ICRD (Item 5 in Fig. 2) before applying the
the residual performance of the buildings becomes critical when weights. A series of displacement transducers are installed at pre-
considering safety of rescue operations. From a longer term per- designed locations. Then, the weights are slowly applied on the
spective, it would provide evidence to guide decisions as to repair slab symmetrically. The external weights in D-0.91 and D-1.16
or rebuild damaged buildings. However, little studies have been are 9 ton and 12 ton, respectively. Moreover, several steel plates
carried out on the residual behavior of buildings following sudden are also placed at the slab extensions (hatched area in Fig. 1) to
column removal. Thus, in this study, a series of three slab-beam- simulate rotational constraints from the slab in surrounding panels
column specimens were designed and tested. A specimen, which partially. The details of ICRD are illustrated in Fig. 3. As shown in
 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184 177

R3 @60

R3 @60
R3 @60
(6)R3 @120

(46)R3 @50
(6)R3 @120
(46)R3 @50

(6)R3 @120

R3 @60

R3 @60

R3 @60
(46)R3 @50 (46)R3 @50

(46)R3 @50

(29)R3 @80
R3 @60 R3 @60 1R8
R3 @60 2R6 T10 T10

R3@ 60 R3@ 60
2R8 2R8

T10
3R6 T13 T13

R3@ 60 R3@ 60
2T10 2T10
R3 @60 R3 @60 R3 @60

Fig. 1. Dimensions and Reinforcement Details of Typical Dynamic Specimen D-0.91 (in mm).

the figure, the ICRD comprises a load cell, a pin support, a steel col-
Table 1 umn, and a hemisphere tip. The special designed ICRD is not really
Dimensions of structural components. removed during the dynamic tests, but rather rotates thereby
releasing the axial force after horizontal impact. To release the
Dimensions Prototype Frame Model Frame
(mm) (mm) axial force in ICRD with the required speed, steel plates with
weights of 190 kg were hung as a pendulum for impacting (refer
Beam Section Interior Beam 305
552 80
138
Perimeter Beam 305
451 80
112 to Item 6 in Fig. 2).
Beam Span Interior Beam 9000 2250 As mentioned above, the Dynamic Specimen D-0.91 and D-1.16
Perimeter Beam 9000 2250 will be re-tested in a push-down loading regime after vibration and
Column Removed Column: 500
500 125
125 stable in position to capture their residual behavior. Fig. 4 illus-
Section B4
trates the test setup for Specimen R-0.91. As the test setup for
Remaining 500
500 170
170
Columns Specimen R-0.91 is similar to that for control specimen, which
Slab Thickness 254 64 has been described in detail in authors’ previous paper (Qian
Concrete Cover of Slabs 25 7 et al. [12]), only a brief description is given in this section. Before
Concrete Cover of Beams 40 10
conducting subsequent tests, the notionally removed column is
moved away from the original position. To simulate uniform dis-

Table 2
Reinforcement details.

Reinforcement Ratio Prototype Frame Model Frame

Beam Longitudinal Rebar Interior Beam Positive Moment Top 0.7% 0.85%
Bottom 1.6% 1.6%
Negative Moment Top 3.1% 3.4%
Bottom 1.6% 1.6%
Perimeter Beam Positive Moment Top 0.5% 0.7%
Bottom 1.3% 1.3%
Negative Moment Top 2.8% 2.7%
Bottom 1.3% 1.3%
Beam Transverse Rebar Interior Beam Beam End 0.28% 0.38%
Mid-Beam 0.42% 0.38%
Perimeter Beam Beam End 0.37% 0.38%
Mid-Beam 0.37% 0.38%
178 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184

Table 3
Reinforcement properties.

Reinforcement Diameter (mm) Yield strength (MPa) Ultimate strength (MPa) Yield strain (106) Ratio of elongation
R3 3 358 439 2014 10.9%
R6 6 445 561 2190 17.5%
R8 8 450 556 2474 16.1%
T10 10 503 635 2609 15.1%
T13 13 473 592 2343 16.9%

1) Weights
2) Specimen 1) Hydraulic jack
3) Displacement transducer
1 2) Steel assembly
1 4) Accelerometer 3) Load distribution tree
4) Displacement transducer
5) Instantaneous Column
5) Specimen
Removal Device (ICRD)
6) Steel supports
6) Pendulum for impact on
ICRD 2
7) Steel support
3 3
4

6 4 2
5 5

7 6

Fig. 2. Specimen D-0.91 ready for test. Fig. 4. Specimen R-0.91 Ready for Test.

2.3. Instrumentation

In dynamic tests, a series of linear variable deformation trans-

ducers (LVDTs), load cells, and strain gauges are installed both
internally and externally to monitor the variation of displacement
and local behavior of the specimens. Due to the limited channel
numbers in Dynamic Strain Meter, only one load cell, six LVDTs
and three strain gauges are measured in dynamic tests. The loca-
tions of LVDTs, strain gauges are given in Fig. 6. The load cell, which
is located at the ICRD, as shown in Fig. 3, is activated during
dynamic tests. In subsequent tests, similar to the tests for control
specimen (Qian et al. [17]), the LVDTs, load cells, and strain gauges
are recorded by Static Strain Meter, rather than Dynamic one. Fur-
thermore, in this phase, the load cell beneath the hydraulic jack
(Item 1 in Fig. 4) is activated to record the force applied on the slab.
It should be noted that the sampling frequencies of the Dynamic
Strain Meter is 2000 Hz.

Fig. 3. Details of Instantaneous Column Removal Device or Item 5 in Fig. 2.

3. Experimental results

Three one-quarter scaled specimens with identical dimensions

tributed pressure (dead load or live load in reality) applied on the
slab, a special designed loading distribution tree (Item 3 in Fig. 4),
which has proven to be a reliable means of equivalent applying
uniform distributed pressure on the slab (Dat and Tan [18]; Bailey
et al. [19]), is utilized in this study for residual strength evaluation.
The details of loading distribution tree are illustrated in Fig. 5. Sim-
ply put, the function of loading distribution tree is to distribute the
concentrated force from the hydraulic jack into 12 equally-loaded
points. To ensure the force distribution is equal even at stages of
large displacement, socket joints are designed to connect the trian-
gular steel plates and the steel beams (Dat and Tan [18]; Bailey
et al. [19]). Similar to the dynamic tests, fixed boundary conditions
are applied on the perimeter columns by strong steel supports.
and reinforcement details are tested under push-down loading
regime or simulated sudden column removal scenarios. The main
results from dynamic tests included displacement responses and
the local strain gauge readings. For subsequent tests, the vertical
load-displacement behavior is mainly measured and thus, the stiff-
ness and strength degradation caused by different dynamic dam-
age is also evaluated.
3.1. Quasi-static behavior of control specimen Con-1
3.1.1. Load-displacement curve
As shown in Fig. 7, the yield strength of Con-1 is 165 kN, which
is attained at a vertical displacement of 40.4 mm. The first peak
 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184 179

P 0.5P

0.25P

0.083P

Fig. 5. Details of Loading Distribution Tree or Item 3 in Fig. 4.

G2

Accelerometer

LVDT
Strain Gauge

G1 S1 I1 S2

E1 E2 G3

E3 H1
Fig. 6. The Location of Instrumentation Devices and Strain Gauges.

load of 195 kN is measured at the vertical displacement of 103 mm.
re-ascending of the load resisting capacity is observed when the
displacement increased beyond 178 mm. The ultimate load resist-
ing capacity of the specimen is measured at the displacement of
531 mm. When the displacement reached 567 mm, fracture
occurred in the beam top longitudinal reinforcement nearby the
side column A4 and the load resisting capacity decreased
significantly.
3.1.2. Failure mode
Fig. 8 shows the failure mode of Control Specimen Con-1. Severe
cracks are observed around the perimeter of the slab and beam-
column interfaces. Plastic hinges are formed at the longitudinal
and transverse beam ends (near A4, C4, and B3). Severe concrete
crushing is also observed at the beam ends and along the slab diag-
onal. In addition, the severe torsional failure is observed at the
edge beam A3-A4. However, it is much milder in the interior beam
300
250
FPL TCA
Load (kN)
C3-C4 as the slab extension with applied design service load could
help to resist partial of the tensile force from the slab top reinforce-
ments. As shown in Fig. 9, outward movement (negative value) is
initially observed with increasing vertical displacement. When
the vertical displacement reaches 168 mm, inward movement is
measured in the support. Although the extent of movement is
dependent on the stiffness of the steel supports, however, the sign
of movement could indicate the development of compressive arch
action and catenary action in the beams indirectly.
3.2. Dynamic behavior of specimens D-0.91 and D-1.16
As dynamic tests on large RC floor subjected to sudden column
missing scenario are relatively rare, the measured dynamic
response including displacement and local strain reading are given.
3.2.1. The history of reaction force in ICRD
Fig. 10 shows the comparison of the variation of vertical force in
ICRD. Before removal of the support, the measured reaction force in
D-0.91 and D-1.16 are 26.5 kN and 33.2 kN, respectively. The
theoretical reaction force in D-0.91 and D-1.16 are 26.5 kN and
34.0 kN, respectively. The test results indicated that the duration

200 YL of column removal in D-0.91 and D-1.16 are 0.012 s and 0.008 s,
respectively. It is fast enough to simulate column removal due to
150
accidental impact or terrorist attack as the duration is less than
100 1/10 of the natural period of the free vibration of the specimens
Con-1-static (Qian and Li [12]; DoD [20]; Iribarren et al. [21]).
50 R-0.91-static
R-1.16-static
0 3.2.2. Specimen D-0.91
0 150 300 450 600 750 Fig. 11 illustrates the displacement distribution in the speci-
Displacement (mm) men. The maximum vertical displacement reached at measuring
point E2 is 42.5 mm. After vibration, the permanent displacement
Fig. 7. Comparison of the load-displacement curve of Con-1, R-0.91, and R-1.16. at E2 is 40.1 mm. Compared with Con-1, which is tested by push-
180 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184

Torsional failure

Fracture of beam longitudinal rebar

Outward Inward

Fig. 8. The failure mode of Con-1.

Horizontal displacement (mm)

12.0 Time (s)

10.0 0 0.5 1 1.5 2 2.5 3
8.0 0
E1
Displacement (mm)
6.0 -10 E2
4.0 E3
2.0 -20 S1
0.0 S2
-30 I1
-2.0 Con-1-H1
-4.0 R-0.91-H1 -40
R-1.16-H1
-6.0
0 150 300 450 600 750 -50
Vertical displacement (mm) Fig. 11. Vertical displacement distribution in D-0.91.

Fig. 9. Comparison of the horizontal movement of the steel support for column C4. ilar to cantilever beam after sudden removal of the column. Similar
deformation shape of the beams and slab is observed in static test
of Con-1.
5 The crack pattern of D-0.91 is shown in Fig. 12. As shown in the
0 figure, from the top view, wide cracks are formed around the
perimeter of the slab. Fig. 13a shows the strain gauge results of
Reaction Force (kN)

-5
D-0.91. As shown in the figure, yield strain is reached in G1 and
F

-10 G2 but not G3. It should be noted that the locations of the strain
-15 gauges are shown in Fig. 6. The peak strain of G1, G2, and G3 are
5789 le, 3408 le, and 2326 le, respectively.
-20
-25 3.2.3. Specimen D-1.16
-26.5 kN
The displacement distribution in Specimen D-1.16 is shown in
-30 --33.2 kN D-1.16 Fig. 14. Immediately following sudden removal of the column, the
D-0.91
-35
0.0 0.2 0.4 0.6 0.8 1.0
Time (s)
Fig. 10. Recorded history of reaction force in the notionally removed column.

down method, the permanent displacement of D-0.91 is very close

to the displacement at yield strength of Con-1. It is predicted that
part of the steel reinforcements in D-0.91 is yielded. As the dis-
placement at E2 of Con-1 is 16.5 mm when the load is 106 kN,
the Displacement Amplification Factor (DAF), which is defined as
the ratio of peak displacement from dynamic test to that from sta-
tic test, is 2.5. Moreover, the peak displacement from E1 and E3 are
27.2 mm and 26.5 mm, respectively. Thus, the external beams A4-
B4 and B4-C4 are in double curvature. However, the peak displace-
ment from I1 is 20.5 mm and thus, the interior beam B3-B4 is sim-
Fig. 12. Crack pattern of dynamic specimen D-0.91.
 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184 181

6000 25000

5000 20000

4000

Strain ( με )
Strain ( με )

15000
3000
10000
2000

D-0.91-G1 5000 D-1.16-G1

1000
D-0.91-G2 D-1.16-G2
D-0.91-G3 D-1.16-G3
0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
Time (s) Time (s)
(a) (b)
Fig. 13. Local strain of specific reinforcements: (a) D-0.91, (b) D-1.16BI1SC1.

measured maximum displacement at the measuring points E1, E2,

E3, S1, I2, S2 are 119 mm, 159 mm, 124 mm, 73 mm, 80 mm, Time (s)
69 mm, respectively. As at the load of 136 kN, the measure displace- 0 1 2 3 4
ment in E2 of Con-1 is 35.6 mm, DAF in this specimen is 4.5. More-

Vertical Displacement (mm)

over, different from D-0.91, the deformation of the specimen keeps
increasing after vibration. Half an hour later, the displacement in
E1, E2, E3, S1, I2, and S2 increases to 177 mm, 249 mm, 184 mm,
108 mm, 127 mm, and 99 mm, respectively. Thus, similar to creep,
the vertical displacement at E2 increased by 56%. As the phe-
nomenon is first raised by the dynamic tests for progressive col-
lapse, it is worth to explain well. The authors postulate that the
developing of plastic strain in the reinforcements of D-1.16 has
not finished when the specimen reaches its first peak displacement.
Further development of plastic strain in the slab and beam rein-
forcement will result in the change of stiffness and bring in further
load redistribution. The vertical movement cannot stop until the
new balance can be finally achieved. More dynamic tests with pre-
cise instrumentations should be carried out to deeper understand-
ing of the phenomenon. The crack pattern of D-1.16 is shown in
Fig. 15. As shown in the figure, considerable torsional failure is
formed in the edge beam A3-B3, which is similar to the failure mode
of Control Specimen Con-1. As shown in Fig. 13b, all strain gauges
G1, G2, and G3 are yielded and the peak strain of G1, G2, and G3
are 22,560 le, 12,779 le, and 12,102 le, respectively.
3.3. Residual behavior of specimens R-0.91 and R-1.16
As mentioned above, after dynamic tests, the weights are
removed and the specimens are rebounded. Then, subsequent test
with similar setup as Con-1 is carried out, as shown in Fig. 2.
3.3.1. Specimen R-0.91
The load-displacement curve of R-0.91 is shown in Fig. 7. As
shown in the figure, R-0.91 has residual deformation of 12 mm in
E2 after removal of the weights. The initial stiffness of R-0.91 is
2.8 kN/mm, which is about 68% of that of Con-1. Moreover, there
is no evident first peak load observed in R-0.91. Compared to
Con-1, R-0.91 developed less compressive arch action and com-
pressive membrane action. However, the behavior of R-0.91 in
large deformation stage is similar to that of Con-1. The ultimate
capacity of R-0.91 is 252 kN, which is about 105% of that in Con-
1. Furthermore, the ultimate displacement in R-0.91 is 600 mm
while that in Con-1 is 566.9 mm. The larger ultimate displacement
in R-0.91 may be because of the dynamic tests achieved smoother
deformation and which delayed the fracture of reinforcements.
Moreover, the dynamic vibration had loosened the anchorage of
0
E1
-40 E2
E3
-80 S1
S2
-120 I1
-160
-200
Fig. 14. Recorded history of the vertical displacement of the notionally removed
column in D-1.16.
the reinforcements. As shown in Fig. 9, comparing to Con-1, less
outward movement but similar inward movement is recorded in
the steel support of R-0.91. The failure mode and crack pattern of
R-0.91 is shown in Fig. 16. In general, it is very similar to that of
Con-1. Detail discussion will be given in the section below.
3.3.2. Specimen R-1.16
As shown in Fig. 7, the residual displacement in E2 of R-1.16 is
190 mm after removal of the weights. As severe concrete crushing
has occurred in the slab and beam ends, compressive arch action
and compressive membrane action have vanished after dynamic
test. When re-loading by push-down loading regime, the load
resisting capacity is mainly attributed into the tensile membrane
action and catenary action. For R-1.16, the ultimate load capacity
and ultimate displacement is 243 kN and 679 mm, respectively.
Thus, comparing to Con-1, R-1.16 has similar ultimate load capac-
ity as Con-1 but could achieve even larger ultimate displacement.
As shown in Fig. 9, no obvious outward movement was observed
in this specimen. In contrast, the inward movement just began to
develop once the vertical load applied on the Specimen R-1.16.
4. Discussion of test results
4.1. Reliability of single-degree-of-freedom (SDOF)
SDOF model and energy based model are commonly utilized
(Izzuddin et al. [13]; Qian and Li [14]) to predict the dynamic
behavior of the test specimen if the load resisting function of the
specimen was captured by push-down analysis or tests. Thus, the
182 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184
Torsional failure
Fig. 15. Crack pattern of dynamic specimen D-1.16.
dynamic responses of the specimens are predicted by SDOF model
in this study.
The SDOF model consists of a single mass element that is con-
nected to a rigid support through a linear elastic spring and viscous
dashpot and expressed by Eq. (1). The main parameters in the
model are described briefly as below. For the details of the model,
please refer to Qian and Li [14].
me €x þ ce x_ þ ke x ¼ PðtÞ  RðtÞ
where me is the equivalent mass; ce is the equivalent viscous damp-
ing coefficient; ke is the effective stiffness; PðtÞ and RðtÞ are the
applied force and reaction force at corner joint respectively; x; x_
and €
x are the displacement, velocity and acceleration of the mass,
respectively. The parameters me ; ke ; and ce are expressed as below:
The equivalent mass me is mathematically expressed as follows:
Z capacity of the frames may be controlled by the CAA or CMA. How-
X ever, when sufficient continual reinforcements were installed in
me ¼ mðzÞ½wðzÞ2 dz þ mk ½wðzk Þ2
k
where mðzÞ is the distributed mass function; wðzÞ is the shape func-
tion, mk is the concentrate mass k at location of zk ; wðzk Þ is the shape
function value at point zk .
The equivalent stiffness ke is defined by secant stiffness which is
measured based on the load resisting function measured in the test
from Control Specimen Con-1.
The equivalent damping coefficient ce is determined by Eq. (3).
Thus, it is varying with the changing of the equivalent stiffness. The
value of damping ratio f is assumed to 5%. For the detail of the
models, please refer to
pffiffiffiffiffiffiffiffiffiffiffiffi
ce ¼ f
2 K e me
As shown in Fig. 17, the predicted maximum displacement at E2
is 38 mm and 156 mm for Specimens D-0.91 and D-1.16, respec-
tively. Compared to test results, SDOF predict the maximum dis-
placement very well. However, the predicted residual
displacement at E2 is 18 mm and 106 mm for Specimens D-0.91
and D-1.16, respectively. They are much lower than the measured
ones. Thus, SDOF will over-estimate the vibration of the system
and provide unsafe results regarding the residual displacement.
4.2. Stiffness and strength degradation
As the pressure applied on D-0.91 is less than the yield strength
of Con-1, comparing to Con-1, the stiffness and strength degrada-
tion of R-0.91 is insignificant. The slight degradation of the stiffness
and strength in R-0.91 is mainly due to the dynamic effects ampli-
fied the applied pressure, which result in D-0.91 experiencing
some extent of inelastic behavior. However, for R-1.16, as severe
plastic deformation is concentrated in the specimen during the
dynamic test, the stiffness and strength degradation is mainly con-
trolled by the tensile catenary action (TCA) or tensile membrane
action (TMA) in large deformation stage. As considerable TCA and
TMA (even larger than the first peak load due to strain hardening
and CAA and CMA) could develop in the Con-1, the stiffness and
strength degradation in R-1.16 is also limited. As shown in
Fig. 18, the structures may have different types of load-
ð1Þ displacement curves when different scenarios of missing column
are considered. For example, the frame subjected to the missing
of a corner column, flexural strength may dominate the load resist-
ing capacity. When the frame subjected to the missing of a middle
or interior column, the sufficient horizontal constraints may allow
considerable compressive arch or membrane action to develop. If
the continuity of the reinforcements is not well, the load resisting
ð2Þ
the beams and slabs, significant TCA or TMA may be able to
develop and they may achieve larger load resistance than CAA
and CMA. Then, the ultimate loading resisting capacity of the spec-
imens is controlled by the TMA and TCA. The authors believed that
the stiffness and strength degradation in the frame with different
controlling load resisting mechanisms may behave differently. As
little tests had been carried out, it is worth to conduct more tests
for specimens subjected to different scenarios of missing columns
in the future.
4.3. Effects on failure modes and load resisting mechanisms
ð3Þ
4.3.1. Failure modes
As shown in Fig. 16a–c, similar crack pattern is observed in Con-
trol Specimen Con-1 and Dynamic Specimen D-0.91 and D-1.16.
Especially for D-1.16, similar to Con-1, torsional rotation is obvi-
ously observed in the edge beam A3-A4. Thus, the push-down
method suggested in this study (quasi-statically increasing the
equivalent pressure) could achieve similar crack pattern and fail-
ure modes of the specimens subjected to simulated sudden column
missing scenario. The proposed push-down method could be a reli-
able alternate method to equivalently investigate the dynamic
response of RC floors when the data pool of dynamic increase fac-
tor (DIF) is built reliably. Comparison of the Fig. 16a–e, it is found
that Specimens R-0.91 and R-1.16 could develop similar failure
mode as Control Specimen Con-1. However, it is found that the
deformation in Specimens R-0.91 and R-1.16 is smoother. This is
due to the weights applied on the slab in dynamic tests being clo-
ser to uniform distributed pressure than the multi-points loading
in Con-1.
 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184 183

Torsional Hinge

Concrete Crushing
Plastic Hinge
Plastic Hinge

(a) (b)

Torsional Hinge Torsional Hinge

Concrete Crushing

Plastic Hinge Plastic Hinge

(c) (d)

Torsional Hinge

Concrete Crushing

Plastic Hinge

(e)
Fig. 16. Comparison of crack pattern and failure modes: (a) Con-1, (b) D-0.91, (c) D-1.16, (d) R-0.91, (e) R-1.16.

4.3.2. Load resisting mechanisms 5. Concluding remarks

As stated in (Qian and Li [17]), considerable CAA and CMA is
developed in Control Specimen Con-1 in relatively small deforma-
tion stage and significant TCA and TMA are developed in the beams
and slab in large deformation stage. However, comparing to Con-1,
the load resistance from CAA and CMA of R-0.91 is less than that of
Con-1. However, similar TCA and TMA are developed in R-0.91
comparing to Con-1. For Damaged Specimen R-1.16, no CAA and
CMA are developed as the dynamic tests had reached severe con-
crete cracks and crushing in the beam and slab. However, similar
to R-0.91, similar TCA and TMA are developed in R-1.16 in large
deformation stage. Thus, comparing to CAA and CMA, TCA and
TMA are more robust for residual behavior.
A series of dynamic tests were conducted first to study the
dynamic behavior of multi-bay beam-column-slab substructures
subjected to the sudden column missing scenario. As the substruc-
tures could survive after sudden removal of the column, their
residual behavior was also captured by conducted subsequent
push-down tests by applying multi-points loads on the slab. As
the designed Instantaneous Column Removal Devices (ICRD) has
been proved satisfy the requirements of axial force release time,
the test setup could capture the dynamic behavior of the speci-
mens well. Due to the help of transverse beams and RC slab, both
specimens with service load of 0.91(DL + LL) and 1.16(DL + LL)
184 K. Qian, B. Li / Engineering Structures 148 (2017) 175–184
Fig. 17. Comparison of the measured and predicted displacement response.
300
250
200
Load (kN)
150
100
Flexural Actioin Dominated
50 CAA or CMA Dominated
TCA or TMA Dominated
0 collapse resistance of reinforced concrete beam column sub-assemblages.
0 150 300 450
Displacement (mm)
Fig. 18. Typical shape of load-displacement curve for RC frames under missing
column cases.
could survive after the sudden removal of a penultimate exterior
column. However, the stability of Specimen D-0.91 is attributed
to the extra load resisting capacity from CAA and CMA. The survival
of Specimen D-1.16 is mainly attributed to considerable TCA and
TMA developed in the beams and slab, respectively.
The displacement amplification factor (DAF) of Specimen D-
0.91 and D-1.16 is 2.5 and 4.5, respectively. Larger DAF is obtained
in D-1.16 due to its more severe plastic damage during dynamic
tests. It is found that D-0.91 refers to the point of termination of
the dynamic vibration due to CAA and CMA and thus, relatively lit-
tle plastic deformation. D-1.160 s movement is stopped attributed
to the load resistance from TCA and TMA in large deformation
stage. It is predictable that DAF will close to 2.0 if lower service
load is applied and the specimen only has elastic behavior.
Subsequent push-down tests indicated that the initial damage
due to dynamic test will degrade the first peak strength and initial
stiffness of the Specimen R-0.91 significantly. This is mainly
because of the initial damage from dynamic test had crushed par-
tial of the concrete with large compressive stress and thus, the con-
tribution of CAA and CMA in load resisting capacity is decreased.
For Specimen R-1.16, no CAA and CMA were developed because
of the dynamic tests result in severe concrete crushing and spalling
in the beam ends. However, the initial damage will not affect the
development of TCA and TMA in both specimens significantly.
Thus, comparing to CAA and CMA, TCA and TMA are more reliable
for residual strength of the structures subjected to the loss of an
View publication stats
exterior column scenario. Similar failure modes and deflection pro-
file are observed subsequent specimens R-0.91, R-1.16, and the
control specimen Con-1, although the subsequent specimens
achieved smoother deformation in the slab.
SDOF models are employed to predict the dynamic displace-
ment of the specimens in dynamic phase. It was found that SDOF
could predict the peak dynamic displacement well while it may
over-estimate the vibration and under-estimate the permanent
residual deformation.
Acknowledgements
The financial assistance provided by the Defence Science &
Technology Agency, Singapore and National Nature Science Foun-
dation of China (Grant Nos: 51568004, 51478118, and
51678164). The authors wish to express their gratitude for the sup-
port received.
References
[1] Sasani M, Kropelnicki J. Progressive collapse analysis of an RC structure. Struct
Des Tall Spec Build 2008;17(4):757–71.
[2] Sasani M, Bazan M, Sagiroglu S. Experimental and analytical progressive
collapse evaluation of actual reinforced concrete structure. ACI Struct J
2007;104(6):731–9.
[3] Su YP, Tian Y, Song XS. Progressive collapse resistance of axially-restrained
frame beams. ACI Struct J 2009;106(5):600–7.
[4] Qian K, Li B. Slab effects on response of reinforced concrete substructures after
loss of corner column. ACI Struct J 2012;109(6):845–55.
[5] Qian K, Li B. Experimental study of drop-panel effects on response of reinforced
concrete flat slabs after loss of corner column. ACI Struct J 2013;110
(2):319–29.
[6] Yu J, Tan KH. Experimental and numerical investigation on progressive
Eng Struct 2013;55:90–106.
[7] Ren PQ, Li Y, Lu XZ, Guan H, Zhou YL. Experimental investigation of progressive
collapse resistance of one-way reinforced concrete beam-slab substructures
under a middle-column-removal scenario. Eng Struct 2016;118:28–40.
[8] Bao YH, Kunnath SK. Simplified progressive collapse simulation of RC
framewall structures. Eng Struct 2010;32(10):3153–62.
[9] Li Y, Lu XZ, Guan H, Ye LP. An improved tie force method for progressive
collapse resistance design of reinforced concrete frame structures. Eng Struct
2011;33(10):2931–42.
[10] Shan SD, Li S, Xu SY, Xie LL. Experimental study on the progressive collapse
performance of RC frames with infill walls. Eng Struct 2016;111:80–92.
[11] Sheffield C, Audrey K, Hoon VNP. An instrumented full-scale building
progressive collapse test. Proc. Of 14th International Symposium on
Interaction of the effects of munitions with structures. Seattle, Washington;
2011.
[12] Qian K, Li B. Dynamic performance of RC beam-column substructures under
the scenario of the loss of a corner column—Experimental results. Eng Struct
2012;42:154–67.
[13] Izzuddin BA, Vlassis AG, Elghazouli AY, Nethercot DA. Progressive collapse of
multi-storey buildings due to sudden column loss—Part Ⅰ: simplified
assessment framework. Eng Struct 2008;30(5):1308–18.
[14] Qian K, Li B. Quantification of slab influences on the dynamic performance of
RC frames against progressive collapse. J Perform Constr Facil 2015;29
(1):4014021–29.
[15] CP 65. Structural use of concrete, Part 1. code of practice for design and
construction. Singapore Standard; 1999.
[16] ACI Committee 318. Building code requirements for structural concrete (ACI
318-08) and commentary (318R-08). Farmington Hills, MI: American Concrete
Institute; 2008. pp. 433.
[17] Qian K, Li B, Zhang ZW. Influence of multicolumn removal on the behavior of
RC floors. J Struct Eng 2016. http://dx.doi.org/10.1061/(ASCE)ST.1943-
541X.0001461, 04016006.
[18] Dat PX, Tan KH. Experimental study of beam-slab substructures subjected to a
penultimate-internal column loss. Eng Struct 2013;55:2–15.
[19] Bailey CG, Toh WS, Chan BM. Simplified and advanced analysis of membrane
action of concrete slabs. ACI Struct J 2008;105(1):30–40.
[20] DoD. Design of building to resist progressive collapse. Unified Facility Criteria,
UFC 4–023-03. Washington (DC): US Department of Defense; 2009.
[21] Iribarren BS, Berke P, Bouillard P, Vantomme J, Massart TJ. Investigation of the
influence of design and material parameters in the progressive collapse
analysis of RC structures. Eng Struct 2011;33(10):2805–20.